Effects of Organizational Structure on the Behavior and Performance of Polar and Space Work Teams*

ABSTRACT

Despite its importance, there is little sociological research on optimization of behavior and performance of work teams in extreme environments. Addressing the need identified by a growing number of social and behavioral researchers, that researchers study the structure and processes of such work teams, the authors explore the effects of crew size, mission duration, mission phase, and crew heterogeneity on rates of "off-nominal" or deviant behavior in a sample of polar expeditions and space missions. Surprisingly, the data from this pilot study indicates that larger, more heterogeneous crews actually have lower rates of off-nominal behavior than smaller more homogenous crews. The data from all cases provide support for the anecdotal "third-quarter phenomenon," that behavioral problems peak in the third quarter of missions, in homogenous crews. However, when the Apollo space missions are eliminated, this effect is substantially diminished. The implications of this effect-specification, and of the other results for the larger project are discussed.

Surprisingly little sociological research has been done on groups and organizations operating in extreme environments. This appears to be the result of several factors.

First, extreme environments (e.g., undersea, aerospace, and polar) are settings so hostile to life that emphasis has often been placed on the technology that attains mission goals and ensures crew survival. What attention has been paid to human factors has often been on the human-equipment interface, as in cockpit design. Other attention has been paid to the need to "select out" clear cases of psychopathology (e.g., the Navy’s desire to eliminate the incidence of the "screaming seaman" phenomenon aboard submarines or the aerospace industry’s wish to keep suicidal pilots out of the cockpit).

Second, the public’s view that the men and women who live and work in extreme environments are heroes, and heroes do not behave badly or conduct themselves in any other than an optimal way, has also militated against such research. This view derives, in part, from a tradition of military management and professionalism from early extreme environmental exploration and the combat setting and has carried over into those organizations that employ civilians and/or military personnel to carry out extreme environmental work. For example, among space flyers, a kind of machismo creed prevails. It is anathema to suggest that they experience behavioral and social difficulties in their lives, in training, or on missions. Examination of space mission records shows just how difficult this creed makes it to determine if a space flyer is ill, depressed, or upset. The way they express their distress is a study in minimalism -- a case of diarrhea, for instance, may be expressed as "stomach awareness." Furthermore, since groups of them often feel it is a matter of honor to cover up for one another, it is difficult to ascertain how well they are functioning together from their demeanor or self-reports.

Third, the perception on the part of many policymakers that sociology, unlike clinical and experimental psychology, or biomedical studies, is not a valid or scientific approach to the study of group performance, has also hindered sociology’s access to the study of these groups.¹

Finally, each of these factors have contributed to the lack of quantitative or quantifiable sociological data on the behavior and performance of extreme environmental work teams, and this lack of data has, in turn, reduced the interests of sociologists in studying such phenomena.

However, as humans increasingly go to work in extreme environments, these factors are fast receding in the face of more numerous situations demanding studies of social components, the inability for the military management and professional model to effectively address these situations, and the fading of the heroic aspect of this kind of work as more people engage in it. In recent years, researchers in psychiatry and in psychology have been among the first to call for a look at the social componenets of extreme environmental work. Their call comes from the accumulating evidence that groups working in extreme environments have widely varying degrees of behavioral and performance difficulties. The first impulse has been to hunt for personality traits that would make people work together well on extreme environmental teams, but they have occasionally raised questions about organizational structure (Sandal 1999). It is increasingly clear that groups and teams in extreme environments organizations need to be studied sociologically.

This pilot study, and the larger project it is a part of, are modest steps in meeting that need. Such research not only holds the promise of developing practical knowledge that may assist in the planning and organization of such missions and expeditions, but may also enrich sociology and sociological theory by testing its theorems and insights among such groups and in such milieux.

The authors’ guiding assumption is that examining off-nominal, deviant, and dysfunctional behavior and performance in the extreme environment is central to optimizing behavior and performance there. This is consistent with Ben-Yehuda’s observation that deviance plays a crucial role in groups, "and a better understanding of the nondeviant can be achieved by understanding the deviant (1985: 216-217)." The nature of sociological deviance, however, is that it is tied to history and locale. Just what is off-nominal, deviant behavior and performance in the extreme environment?

To answer that question, the investigators developed a coding protocol and then tested and refined it using coders with a variety of backgrounds: sociologists and graduate students with no familiarity with extreme environments issues, scientists who dealt with extreme environments issues, and explorers (Dudley-Rowley 1997). Although coders came from various backgrounds, their responses were highly reliable and were used to develop objective principles for identifying deviant behavior and off-nominal performance among workers in extreme environments. The tests involved presenting selected passages from expeditionary and space mission narratives (i.e., logs, diaries, and other accounts), which may or may not have contained incidents expected to be seen as off-nominal performance or deviant behavior. The protocol was presented to multiple, independent coders who recorded any behaviors or events that they deemed deviant or dysfunctional in the passages they read. The respondents did not communicate with one another.

The actions identified by coders fell into three general categories: (1) unusual, bizarre, or puzzling behaviors, such as withdrawal and life-threatening acts, (2) acts of aggression, verbal, and physical, and (3) acts of deliberation, such as resource theft, hoarding or hogging resources, not doing one’s work, and violating safety rules. The final coding protocol enabled the development of quantitative data from existing expedition and mission records for a total of ten polar expeditions and space missions. The specific polar cases are: the 1881-84 Lady Franklin Bay Expedition (LFB), the 1911 Western Party Field Trip of the British Antarctic Expedition (the Terra Nova Expedition [Ter. Nov.]), the 1921-23 Wrangel Island Expedition (Wrangel), the 1929 Dominion Explorers’ Expedition (DomEx), the 1959-60 Field Traverse from Byrd Station to Amundsen Sea and Return of the International Geophysical Year Expedition (IGY), the 1982-84 Frozen Sea Expedition (Frozen Sea), and the 1989-90 International Trans-Antarctica Expedtion (Steger). There are 3 Arctic expeditions of this number, (LFB, Wrangel, and DomEx). Four are Antarctic expeditions (Ter. Nov., IGY, Frozen Sea, and Steger). The space missions are: the 1969 Apollo 11 mission, the 1970 Apollo 13 mission, and the 1982 Salyut 7 mission.

Initial data from the pilot study allowed preliminary testing of a number of a priori structural hypotheses concerning rates of deviance among crews in extreme environments. They also provided an opportunity to empirically evaluate the anecdotally-noted "third-quarter phenomenon" -- the notion that crew conflict and deviance will peak just after mission mid-point -- as well as the opportunity to develop empirically-based hypotheses that can be tested in the larger data set ultimately produced by the longer-term project.² We thus may have the opportunity to identify "serendipitous" results, which later can be subjected to more rigorous independent testing.

Although it is worth emphasizing that one must be very cautious in drawing conclusions from such a limited data set, it should be noted that these data provide a new and unique opportunity to explore these issues among such crews and in such environments.

HYPOTHESES

The specific a priori hypotheses considered are, first, from basic arguments of structural sociology (e.g., Blau 1977, Mayhew 1980, Mayhew et al. 1995):

Second, from anecdotal accounts:

Basic arguments of structural sociology posit that larger numbers of people in contact with one another have more opportunities for interpersonal conflict, and increasing heterogeneity of any kind is expected to correlate with increasing opportunities for conflict and deviance (e.g., Blau 1977). That larger numbers of people will generate increasing opportunities for conflict is consistent with a combinatorial mathematical standpoint as well (e.g., Mayhew and Levinger 1976).

The third-quarter, or third-stage phenomenon was, perhaps, first noted by Rohrer (1961). He identified three stages of reaction to prolonged isolation, confinement, and stress. The first stage was a heightened anxiety brought about by the perceived dangers of the situation. The second stage occurred as the crew settled down to a daily routine, and was marked by depression and regrets about having joined the mission. The third stage is a period of anticipation, but features increased emotional outbursts, aggression, and rowdiness (Rohrer 1961). Earls called it a "half-way syndrome" which corresponded with a low point in the crew’s morale beginning halfway through an expedition (Earls 1969). Anecdotally, it has been thought that these phases are present regardless of the length of the mission, whether it be days, weeks, or months. Sheddan thought that with longer duration missions, the phenomenon, however, would be more pronounced (1995). However, the question about the third-quarter phenomenon has remained open because previous studies have been hindered by the lack of quarterly data on missions (Bechtel and Berning 1991).

DATA, MEASURES, AND METHODS

Employing the pretested protocols (Appendix A) multiple coders independently recorded the incidents of deviance they encountered in the written records of the missions and expeditions. Discrepancies and disagreements among coders were resolved by individually interviewing discrepant coders, providing minimal information to resolve disagreements and discrepancies, and recoding. Care had to be taken to avoid biasing coders by forcing them to adopt notions thought deviant or dysfunctional by the investigators. However, the typical discrepancy involved younger coders not reading closely enough. It was not necessary to re-code passages or whole records often.

There was a high degree of inter-coder reliability (r²), averaging .93 for all ten cases.³

After establishing the inter-coder reliability per record, numbers of deviant acts by quarters were determined by averaging over the different coders’ estimates for each quarter. Coders noted the specific mission phase deviant incidents occurred in, and then entered them into the data set containing corresponding information on crew characteristics (sex, nationality, experience, and age), crew size, mission duration, and ancillary data (like mission objectives).

Indices of crew heterogeneity with regard to sex, nationality, and experience were constructed following Blau (1977).⁴ Age heterogeneity was indexed by the range of crew ages. To make rates of deviance comparable across crews of differing sizes and missions of differing length, a standardized measure of deviance was constructed.⁵ From the data, two tables were constructed (Tables 1 and 2). These tables (Table 1: Means for Ten Polar Expeditions and Space Missions) and (Table 2: Data for All Quarters of Ten Polar Expeditions and Space Missions) presented the format to construct the x,y graphs to describe a series of zero-order and first-order relationships essential to testing the hypotheses.

RESULTS

Mission durations ranged from 6 to 1080 days. Crew sizes ranged from about three (2.75) to more than 20 (23.1).⁶ Sex heterogeneity ranged from 0 to 0.45; nationality heterogeneity, 0 to 0.82; age heterogeneity, 0 to 60; experience heterogeneity, 0 to 0.43. There is some independence among the different dimensions of heterogeneity. For instance, although the Steger crew is homogenous with respect to sex and experience, there is considerable heterogeneity with respect to nationality.

Standardized rates of deviance ranged from a low of 0 (IGY) to a high of 220 (Apollo 13). In this regard, Apollo 13, for fairly understandable reasons, is a clear outlier, which will require special attention. The next highest rates of deviance are Apollo 11 (80) and Salyut 7 (60). It is interesting that the highest rates of deviance are found in the space missions, for as the authors noted above, it has been argued that such crews and missions are essentially without deviance. One might argue that the records for space missions are somehow more detailed than those of the polar expeditions, artificially producing high rates of deviance. However, the space missions and polar expeditions were narratized by their authors on about the same scale of reportage. Had the data for the space missions been extracted from the mission transcripts, the space missions and polar expeditions would have been analyzed separately.

Surprisingly, Figure 1 shows that, excluding Apollo 13, rates of deviance appear to generally decline with increasing crew size. Contrary to our expectation, and Hypothesis 1, the average rate of deviance is 70 for the smallest crews (Apollo 11 and Salyut 7), and then, ignoring the spike of 220 for Apollo 13, drops off to 30 for crews of about 4, and then remains at 20 and under for even larger crews, falling to 10 and under for crews with more than seven members. Crews under 3 have an average deviance rate of 70,⁷ while those larger than 3 have an average rate of only 14.3.

The rate of deviance also tends to decline with increasing crew heterogeneity, contrary to Hypothesis 2. Sex-homogenous crews had an average rate of deviance of 56.67, while it was about half that, 30, for those with some sex-heterogeneity. If Apollo 13 is excluded from the first group, the sex-homogenous average is 24, however. Only two crews in our preliminary data set were homogenous with regard to nationality -- Apollo 13 and Apollo 11. Their average rate of deviance was 150. More heterogeneous crews had an average of 20. Dropping Apollo 13 from the first group, produces a rate of 80 represented by the one remaining nationally-homogenous crew. Crews homogenous with respect to experience had an average rate of 78, while those with some heterogeneity had a rate of only 14. Dropping Apollo 13 from the first group, gives an average rate of 42.5.

The standardized mean rate of deviance for all ten missions and expeditions is 46. Without Apollo 13, the standardized mean rate of deviance is 26.67. Including Apollo 13, small groups and homogenous groups were substantially above the mean rate of deviance for ten missions and expeditions.

Our hypothesis that greater heterogeneity would, by chance alone, increase the probability of misunderstandings and miscommunications, and thereby raise the rate of deviance, is contradicted by this evidence, and the results are relatively consistent across the different dimensions of heterogeneity, even experience. A possible explanation, which might be tested in the larger data set, is that such differences create a greater differentiation and interdependence among crew members -- what Durkheim (1893) referred to as organic solidarity (Nolan et al. 1991). Thus, rather than increasing deviance, it might actually retard it. For those less concerned with the sociology, and more concerned with the instrumentality of such groups, we tentatively offer the inference that in crews in extreme environments, "diversity works."

Figure 2 also shows that, contrary to expectation, deviance generally tends to fall, not increase, with increasing mission duration. With Apollo 13 included, the average rate of deviance for missions of less than 90 days is 64 and 28 for those longer. Excluding Apollo 13 and the Salyut 7 blip, the rate is 25 for the short missions and 20 for the long ones. The mean rate of deviance for the 8 missions remaining is 22.5. If only Apollo 13 is dropped, the rates are about equal: 25 and 28 , respectively, with the short missions being just below and the long missions being just above the mean rate of deviance for 9 missions.

Finally, turning attention to the anecdotal "third-quarter phenomenon", our final hypothesis, we find a truly intriguing and unanticipated result. Figure 3 shows that, averaged across all types of crews, and all durations of missions, there is, indeed a third-quarter spike in the rate of deviance. To our knowledge, this is the first quantitative support for this idea. But what is truly intriguing and completely unanticipated about this effect only emerges when we look separately at homogenous and heterogeneous crews. Regardless of which dimension of heterogeneity is used (Figure 4), the same pattern emerges: homogeneous crews clearly have a third-quarter spike, and heterogeneous crews do not.

This may explain the continued anecdotal status of this effect. Those researchers and mission planners who clearly sensed and believed in it may have been looking only at homogenous groups and those who did not may have been looking at more heterogeneous groups. This serendipitous finding certainly warrants more rigorous examination and testing in the larger data set.

Again, however, the investigators urge caution, for only 10 cases are represented here from approximately 50 currently in different stages of processing. When we take out Apollo 13, analyzing for the 9 remaining cases, a substantial second-quarter rate of deviance spike emerges (42.22), with no other quarter being above the mean rate of deviance of 26.67. Analyzing for heterogeneity in terms of sex, nationality, age, and experience, we find crews homogenous for the dimensions are far above the mean rate of deviance in the second quarter (64, 30 , 170, and 90, respectively.).

If we analyze eight missions without either Apollo mission in the analysis, the third-quarter effect dissipates when we examine rates of deviance by mission quarter. However, the third-quarter effect substantially re-emerges in crews homogenous for age and experience. Without the Apollo missions, Salyut 7 is the only age-homogenous crew. There are no nationally homogenous crews without the Apollo missions. Without Apollo 11 and Apollo 13 in the analysis, crews heterogeneous with regard to sex are substantially above the mean, while homogenous crews are substantially below it.

Excluding the Apollo missions, there is no robust evidence of third-quarter phenomenon. However, we may still speculate that the third-quarter effect may be present under two conditions:

(1) where there is immediate and overwhelming threat to life (as in Apollo 13, the mission nearly lost on the way to the moon), and

(2) when crews are relatively homogenous over a number of dimensions.

DISCUSSION AND CONCLUSION

Although most of our a priori hypotheses have been challenged or contradicted by the data for all ten cases in our pilot study, this research has provided new and exciting insights into the effects of social structural factors on deviance in crews in extreme environments. These are, in fact, some of the first quantitative sociological data developed and analyzed on such crews in these settings. That they produced discernable trends over all 10 cases at all is confirmation of the utility of the novel and innovative coding protocols developed for this project.

These tentative and speculative observations suggest that there is something compelling to the structural correlates of polar field expeditions and space missions. The general trend over any of the analytical configurations indicates that heterogeneity over a number of independent dimensions produces lower rates of deviance and dysfunction. Its effect also seems able to outweigh singular events like failure to be retrieved and put in a life-and-death situation, coping with the mental illness of a team-mate, dealing with people who do not want to be on the expedition, deaths, and other egregious circumstances. This suggests that catastrophic events may be dealt with more effectively by crews whose individuals can find appropriate social niches within their crews.

The larger project which follows this pilot study will collect and analyze data from 75 to 100 polar field and space records. More than 50 of which are currently in varying stages of completion. This much larger data set will allow more definitive conclusions and will be able to detect subtler patterns of deviance in these settings.

APPENDIX A

INSTRUCTIONS TO CODERS

Coders were told:

"Deviance, as defined by sociologists, is behavior that is a recognized departure from the expectations and cultural norms that exists within societies, even microsocieties which people the expeditionary settings of the polar regions and space.

A previous round of testing has determined a unifying principle underlying deviance among expeditioners in extreme environments: there are three overlapping general categories of deviant behavior which can be seen in the expeditionary record which cover numerous examples of deviant behavior in extreme environments. The examples given below behind the bullets are meant to be representative, not exhaustive. These are as follow:

I. Unusual, bizarre, or puzzling behaviors, such as:

-- Isolating oneself from the group, becoming withdrawn

-- Suicidal and life-threatening behaviors

-- Paranoid actions

II. Acts of aggression, such as:

-- Verbal aggression, including threats and the language of sexual harrassment

-- Physical aggression, running the gamut from symbolic displays to physical attacks

III. Acts of deliberation, such as:

-- Crewmembers "hogging" the communications systems or making unauthorized calls to Base or Ground

-- Taking things that belong to the group, including resources theft or hoarding

-- Not doing one’s work or botching tasks on purpose

-- Failing to lead or to plan, in the case of those in leader and planner positions

-- Violating safety rules or setting up accidents by deliberate jury rigging

INSTRUCTIONS: The following represent unedited passages from expedition narrative penned by the explorers themselves. In the blank spots provided below each narrative, please number and name each instance of deviance you can discern in it according to the categories named above. There may be many or no instances of deviance as defined by the general categories. And, in this test, I am concerned with deviant behavior solely among the expeditioners themselves, not with any actions on the parts of Ground Control/Base or suppliers."

Coders were then given several raw passages from different expeditions to number and name the incidents of deviance they could discern therein. This is a representative passage from narrative describing the Salyut 7 mission.

"I got up and felt that my mood wasn’t as good as it used to be -- I felt a little sad. It’s natural -- the drab routine has begun.

Something as happened to the water regeneration system. It’s become difficult to cook breakfast. Air is coming up with the water. Food packages have expanded and look like the air bladders of fish. The freeze-dried food doesn’t dissolve very well. We’ll have to wait till the water in the drinking reservoir is finished and then fill it up with good drinking water that doesn’t have any bubbles.

Today Tolia complained to me: "Valentin, how long will I have to keep catching your stuff?" (The sextant, still camera and movie camera are my stuff.)

"Tolia," I told him, "if we start counting what’s yours and what’s mine just one week after the beginning of our mission, it won’t be any good." I could see that he was irritated.

During breakfast we decided to set up shifts for food preparation. Today we conducted a test on solar and terrestrial orientation (SOR). The Ground Control was happy with it even though I wasn’t entirely satisfied. Tonight we summarized the results of our work for the past week. The head of the 19-1 group said that he has worked with all the main crews but has never seen so much done in such a short period of time. Today the doctors told us that we’ve underslept 7 hours and overworked 20 hours. We have to compensate for it somehow. I noticed that after such an evaluation Tolia’s mood improved.

Now it’s time to sleep. The day after tomorrow we have a TV meeting with our families. This is something to look forward to."

REFERENCES

Bechtel, Robert B. and Amy Berning. 1991. "The Third-Quarter Phenomenon: Do People Experience Discomfort After Stress Has Passed?" in From Antarctica to Outer Space: Life in Isolation and Confinement, A.A. Harrison, Y.A. Clearwater, and C.P. McKay (Eds.). New York: Springer-Verlag, 261-266.

Ben-Yehuda, Nachman. 1985. Deviance and Moral Boundaries: Witchcraft, the Occult, Science Fiction, Deviant Sciences, and Scientists. Chicago: University of Chicago Press.

Blau, Peter M. 1977. Inequality and Heterogeneity: A Primitive Theory of Social Structure. New York: Free Press.

Dudley-Rowley, Marilyn. 1997. "Deviance Among Expeditioners: Defining the Off-Nominal Act Through Space and Polar Field Analogs," The Journal of Human Performance in Extreme Environments 2(1): 119-127.

Durkheim, Emile. [1893] 1964. The Division of Labor in Society. New York: Free Press.

Earls, J.H. 1969. "Human Adjustment to an Exotic Environment," Archives of General Psychiatry 20: 117-123.

Mayhew, Bruce H. 1980. "Structuralism vs. Individualism: Part 1, Shadowboxing in the Dark," Social Forces 59: 335-375.

Mayhew, Bruce H., J. Miller McPherson, Thomas Rotolo, and Lynn Smith-Lovin. 1995. "Sex and Race Homogeneity in Naturally Occurring Groups," Social Forces 74: 15-52.

Mayhew, Bruce H. and Roger Levinger. 1976. "Size and the Density of Interaction in Human Aggregates," American Journal of Sociology 81: 1017-1049.

Nolan, Patrick, John Skvoretz, and Gladys Zemo. 1991. "Size, Work Volume, and Differentiation: A Study of U.S. Customhouses," Social Science Quarterly 72: 696-714.

Rohrer, J.H. 1961. "Interpersonal Relationships in Isolated Small Groups" in Psychophysiological Aspects of Space Flight, B.E. Flaherty (Ed.). New York: Columbia University Press: 263-271.

Sandal, Gro M. 1999. "The Effects of Personality and Interpersonal Relations on Crew Performance During Space Simulation Studies," Journal of Human Performance in Extreme Environments, 4(1): 43-50.

Sheddan, Marylin K. 1995. "Role Changes During Long-Term Missions: An Anecdotal Assessment." AIAA.

FOOTNOTES

* Direct all correspondence to Patrick D. Nolan, Department of Sociology, The University of South Carolina, Columbia, SC 29208; pnolan@sc.edu. This research was supported in part by National Science Foundation grants SBR-9729957 and SES-9944042.

¹ For example, the junior author of this report was only judged to be eligible for astronaut candidate training after arguing successfully with space agency administrators that although she was trained in a sociological program, she was an experimental social psychologist as well. Clinical and experimental psychologists are accepted for astronaut candidate training, but sociologists are not.

² This strategy was quite effectively employed by Blau and Schoenherr (1971) in their study of the structural determinants of organizational features. They initially developed and refined a number of hypotheses using data on one set of organizations and then tested them empirically with data on another set of organizations.

³ Inter-coder reliability was calculated through the use of the G-Study, an ANOVA procedure. The model is Y = m + a + b + e where m = overall mean of deviant acts; a = effect of coder; b = effect of mission; and e = random error. The reliability coefficient is expressed as r² = (MSb - MSe) / MSb + (n_i - 1)(MSe). Inter-coder reliability obtained where r² > .75.

⁴ The measure used was Blau's "h" (Blau 1977:9), the probability that two randomly selected crew members would be the same sex, nationality, or have the same number of years of experience in extreme environments.

⁵ DaPrRate = Deviant acts/expedition days/numbers of expedition crew. For readability and ease of presentation, these rates were multiplied by a constant of 1,000.

⁶ Fractional crew sizes result from the fact that in some cases crew members left (e.g., Wrangel) or arrived (e.g., IGY) during the mission or expedition, and crew size had to be averaged over the relevant period.

⁷ The disparity is even larger if Apollo 13 is included; it raises the average for crews 3 and under to 120.